Day 29 - Amazon ElastiCache

Date: 2025-10-16 (Thursday)
Status: “Done”

Lecture Notes

Amazon ElastiCache

Managed in-memory caching service for Redis and Memcached to reduce latency and offload databases.

  • Microsecond reads, Multi-AZ with failover, simple scaling, encryption/auth, automated ops.
  • Redis: rich data structures, backups, replication, cluster mode.
  • Memcached: simple, horizontally scalable cache with auto-discovery.

Common uses: web/mobile acceleration, DB query caching, session stores, leaderboards, pub/sub, queues.

ElastiCache for Redis Features:

  • Data Structures: Strings, lists, sets, sorted sets, hashes, bitmaps, hyperloglogs
  • Persistence: Snapshots and AOF (Append-Only File)
  • Replication: Primary-replica with automatic failover
  • Cluster Mode: Partition data across multiple shards
  • Pub/Sub: Real-time messaging
  • Lua Scripting: Server-side scripting
  • Geospatial: Location-based queries

ElastiCache for Memcached Features:

  • Multi-threaded: Utilize multiple cores
  • Auto Discovery: Automatic node discovery
  • Horizontal Scaling: Add/remove nodes easily
  • Simple: Easy to use, no persistence

Redis vs Memcached:

Feature Redis Memcached
Data Structures Rich (lists, sets, etc.) Simple (key-value)
Persistence Yes No
Replication Yes No
Multi-AZ Yes No
Backup/Restore Yes No
Pub/Sub Yes No
Multi-threaded No Yes

Caching Strategies

Cache-Aside (Lazy Loading)

  • Application checks cache first
  • On miss, load from database and populate cache
  • Pros: Only requested data is cached
  • Cons: Cache miss penalty, stale data possible

Write-Through

  • Write to cache and database simultaneously
  • Pros: Data always fresh, no cache misses on reads
  • Cons: Write penalty, unused data may be cached

Write-Behind (Write-Back)

  • Write to cache immediately, async write to database
  • Pros: Fast writes, reduced database load
  • Cons: Risk of data loss, complexity

Use Cases

Session Store:

# Store user session in Redis
redis.setex(f"session:{user_id}", 3600, session_data)

# Retrieve session
session = redis.get(f"session:{user_id}")

Leaderboard:

# Add score to sorted set
redis.zadd("leaderboard", {user_id: score})

# Get top 10
top_10 = redis.zrevrange("leaderboard", 0, 9, withscores=True)

Rate Limiting:

# Increment counter with expiry
pipe = redis.pipeline()
pipe.incr(f"rate:{user_id}")
pipe.expire(f"rate:{user_id}", 60)
count = pipe.execute()[0]

if count > 100:
    raise RateLimitExceeded()

Hands-On Labs

Lab 43 – AWS Database Migration Service (DMS) (Part 3)

  1. Inspect S3 → 43-12
  2. Create Serverless Migration → 43-13
  3. Create Event Notification → 43-14
  4. Logs → 43-15
  5. Troubleshoot: Memory Pressure → 43-16
  6. Troubleshoot: Table Error → 43-17